From mboxrd@z Thu Jan 1 00:00:00 1970 From: Oliver Neukum Subject: Re: [PATCH torvalds-2.6] cdc_ether: fix autosuspend for mbm devices Date: Mon, 26 Apr 2010 12:30:21 +0200 Message-ID: <201004261230.21713.oneukum@suse.de> References: <201004261214.13836.torgny.johansson@ericsson.com> Mime-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org To: Torgny Johansson Return-path: Received: from cantor.suse.de ([195.135.220.2]:57504 "EHLO mx1.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752766Ab0DZKaX (ORCPT ); Mon, 26 Apr 2010 06:30:23 -0400 In-Reply-To: <201004261214.13836.torgny.johansson@ericsson.com> Sender: netdev-owner@vger.kernel.org List-ID: Am Montag, 26. April 2010 12:14:13 schrieb Torgny Johansson: > Hi! > > In 2.6.33 it seems autosuspend is broken for mbm devices. It is hardly broken. It simply wasn't implemented. > Autosuspend works until you bring the wwan interface up, then the device does not enter autosuspend anymore. > > The following patch fixes the problem by setting the .manage_power field in the mbm_info struct > to the same as in the cdc_info struct (cdc_manager_power). > > I am unsure exactly what that does and why autosuspend doesn't work without it. > Can you guys comment on that? Is this fix the correct approach? The patch is correct. If you use another struct driver_info it is a different driver as far as usbnet is concerned. A driver needs to give some minimal support for autosuspend. mbm didn't do so. > Also, if you like this patch, is it possible to get it included in a minor release of 2.6.33 (e.g. 2.6.33.3) as a bugfix? Up to David. Regards Oliver